Statistical optimization of parametric accelerated failure time model for mapping survival trait loci

Area of Science:

• Genetics
• Biostatistics
• Bioinformatics

Background:

• Standard quantitative trait loci (QTL) mapping methods struggle with skewed survival data and censoring.
• Survival analysis models, particularly the accelerated failure time (AFT) model, are crucial for analyzing time-to-event data.
• Integrating AFT models into QTL interval mapping is essential for studying survival traits.

Purpose of the Study:

• To develop a novel parametric approach for QTL mapping of survival traits.
• To utilize the Expectation-Maximization (EM) algorithm for parameter estimation.
• To employ the Bayesian Information Criterion (BIC) for selecting optimal error distributions and constructing parsimonious mapping models.

Main Methods:

• A parametric QTL mapping approach based on the AFT model.
• Application of the EM algorithm for maximum likelihood estimation.
• Model selection using BIC to identify the best-fitting survival distribution.

Main Results:

• The proposed method was successfully applied to two real datasets.
• Weibull distribution was identified as optimal for mapping heading time in rice.
• Log-logistic distribution was found to be optimal for hyperoxic acute lung injury.

Conclusions:

• The developed parametric approach effectively maps QTL for survival traits, even with skewed distributions and censoring.
• The method provides a robust framework for selecting appropriate survival models in genetic studies.
• Accurate identification of optimal survival distributions enhances the precision of genetic analyses for complex traits.